Measuring Satisfaction

Satisfaction Scales

Many satisfaction specific scales have been used in academic research and in commercial marketing research—there is no established measurement standard such as in the case of length, weight, time, and several other nonpsychological constructs. Traditionally, satisfaction scales have been of the single-item type, which means that one single satisfaction question is asked in the questionnaire. Over time, however, multi-item scales have become more prevalent. They thus require several questions—within the frame of the same questionnaire—about customer satisfaction. An advantage with the multi-item approach is that the reliability of the satisfaction measure can be estimated in terms of the internal consistency of the responses to the individual questions. Such estimates are often obtained by computing Cronbach's alpha, an indicator taking on values ranging from 0 to 1. Values above 0.7 indicate that reliability is at an acceptable level. Indeed, serious academic marketing journals are nowadays demanding that the researcher who is using questionnaires should estimate and report the level of reliability in the measures before he or she proceeds to the analysis, and a multi-item approach offers a convenient way to deal with such demands. The typical approach, then, is to (a) ask several questions about customers' satisfaction with one particular offer, (b) estimate the reliability, and (c) create a new satisfaction variable, for each respondent, to be used in the subsequent analysis (given that the reliability is at an acceptable level). This new variable is usually calculated as the mean response to the individual satisfaction questions in the multi-item measure.

Even though different researchers use different questions and different response formats for their satisfaction measures, one particular multi-item measure has been used more often than others. This measure consists of three questions and was first used in the customer satisfaction barometer developed in

Sweden at the end of the 1980s. The same approach was later used in the American customer satisfaction barometer and in several other national satisfaction barometers. By now, satisfaction data have been collected from a gargantuan number of customers—in many different countries and with regard to many different offers—with these three questions (and the response formats): “How satisfied or dissatisfied are you with XYZ?” (1 = very dissatisfied, 10 = very satisfied), “To what extent does XYZ meet your expectations?” (1 = not at all, 10 = totally), and “Imagine a [the product or offer in focus] that is perfect in every respect. How near or how far from this ideal do you find XYZ?” (1 = very far from, 10 = cannot get any closer). It should be noted that these questions can be employed for two conceptually distinct types of customer satisfaction: transaction-specific satisfaction (i.e., satisfaction after one particular transaction) and accumulated satisfaction (i.e., satisfaction after several transactions).

...